Defining the surgical management of suspected early-stage ovarian cancer by estimating patient numbers through alternative management strategies.

OBJECTIVE

To establish the optimal management strategy for women with suspected stage 1 ovarian cancer.

DESIGN

We created a flowchart to illustrate each of six hypothetical management strategies. These considered two surgical approaches (systematic lymphadenectomy versus no lymph node dissection at all) in combination with three different policies for giving adjuvant chemotherapy.

SETTING

Gynaecological cancer centre, London, UK.

DATA SOURCES: Patient data and published papers.

METHODS

We developed a deterministic model that uses information from multiple sources to estimate patient flow through each level of a hypothesised decision tree.

RESULTS

We estimated that for every 100 cases of suspected early-stage ovarian cancer, there would be 37 cases with 'apparent' stage 1 disease and that of these, two (6%) would be denied potentially life-saving adjuvant treatment if systematic lymphadenectomy was not performed. The number of women given chemotherapy would not, according to our estimates, differ greatly between the two surgical approaches, the 7% increase with systematic lymphadenectomy being because of cases identified as having nodal metastases.

CONCLUSIONS

We present a model of the intraoperative decision-making process that determines the extent of the staging procedure to be performed within our department when early-stage ovarian cancer is suspected. Unless adjuvant chemotherapy is prescribed for all, systematic pelvic and para-aortic node dissection is required to optimise survival. However, in our department, this would result in 32% of women with suspected early-stage ovarian cancer undergoing systematic node dissection. This flexible focused model may facilitate multidisciplinary team discussion when this part of the surgical staging procedure is considered within the context of the population presenting to the team, the morbidity of the procedure within the department and the predictive values of frozen section within that department. As the model is not disease-specific, it may be useful for decision making in other medical disciplines.